← Back to community benchmarks
Qwen3.5-4B
Performance
32k
tokens
317.9
PP tok/s
22.2
TG tok/s
103074
TTFT (ms)
5.9
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
16 GB
GPU Cores
10
Software
oMLX
v0.2.6
macOS
macOS 26.3
Context
32,768
Performance by Context Length
| Context | PP tok/s | TG tok/s | Peak Mem | |
|---|---|---|---|---|
| 1k | 400.7 | 32.3 | 3.9 GB | view |
| 4k | 395.6 | 35.4 | 4.2 GB | view |
| 8k | 387.6 | 32.4 | 4.6 GB | view |
| 16k | 361.8 | 27.6 | 4.8 GB | view |
| 32k | 317.9 | 22.2 | 5.9 GB | current |
| 64k | 242.8 | 14.2 | 8.0 GB | view |
Batching Results
| Batch Size | TG tok/s | Speedup |
|---|---|---|
| 1× | 32.3 | 1.00× |
| 2× | 71.6 | 2.22× |
| 4× | 88.7 | 2.75× |
| 8× | 90.7 | 2.81× |